PROJECT SUMMARY: MODELING PROJECT Our overall aim is to develop, apply, and distribute computational tools that predict the activities for individual enzymes and for multiple enzymes that function together in a biochemical pathway. We integrate concepts and methods derived from comparative protein structure modeling, structure-based ligand docking, integrative structural biology, chemoinformatics, and systems biology. Specific Aim 1. A major focus has been the development and application of structure-based approaches to predict substrates for enzymes of unknown function; when experimental structures are unavailable, we have demonstrated that comparative protein models can be productively used. While we have had substantial success, key challenges remain to enable routine, successful application, including: sampling both covalent and non-covalent intermediates, expanding from simple substrate prediction to predicting activities across a metabolic pathway, and an elementary lack of the right metabolites in our docking libraries. 1a. We will explore a new covalent-docking method to allow sampling both non-covalent intermediates as may be suitable for many enzyme families. 1b. We will broaden our efforts from docking potential substrates against a single target, to doing so against candidate members of an entire pathway; this will be key information to integrative modeling of metabolic pathways. 1c. To address the problem of missing metabolites, we will use chemoinformatics to infer metabolites from the docking of synthetic libraries, which span a much larger chemical space, and have fewer gaps, than the extant metabolite libraries. Specific Aim 2. We aim to develop, apply, and distribute a method for mapping metabolic pathways that will identify the enzymes and ligands in a pathway as well as their order. The goal will be achieved by integrating structural and systems information, such as data from virtual screening, cheminformatics, genomic context, ligand binding experiments, and metabolomics. Specific Aim 3. The Metabolism Project (supported by the Protein Core) will test predictions of metabolite docking on individual enzymes, and predictions of integrative mapping for metabolic pathways, providing crucial feedback to evaluate and improve the methods. The Ligand Discovery Project will provide data for integrative mapping, and structure determination efforts will facilitate comparative protein modeling and, in select cases, test predictions of protein-ligand complex structures generated by metabolite docking. Specific Aim 4. We will make our tools widely used, including, when possible, by non-expert scientists. Thus, we will automate and distribute our open source packages, web servers, databases, benchmarks, and sample applications. These will include optimized libraries of metabolites in dockable forms and an optimized tool for docking these metabolites, a tool for covalent docking, tools within the ZINC family of programs and databases to seek metabolites that resemble synthetic molecules, and to optimize specific libraries, and well as the open source IMP package for integrative pathway.